YT-23266: Fix skip unknown fields

commit_hash:74f3a55474c94110d8a379b1a2d2f75499ea737a

5 files changed, 68 insertions, 28 deletions

diff --git a/yt/yt/core/misc/protobuf_helpers.cpp b/yt/yt/core/misc/protobuf_helpers.cpp
index ff5e4046df..f68eb5de62 100644
--- a/yt/yt/core/misc/protobuf_helpers.cpp
+++ b/yt/yt/core/misc/protobuf_helpers.cpp
@@ -418,7 +418,7 @@ void ToProto(NYT::NProto::TExtensionSet* protoExtensionSet, const TExtensionSet&
     }
 }
 
-void Serialize(const TExtensionSet& extensionSet, NYson::IYsonConsumer* consumer)
+void Serialize(const TExtensionSet& extensionSet, NYson::IYsonConsumer* consumer, const TProtobufParserOptions& parserOptions = {})
 {
     BuildYsonFluently(consumer)
         .DoMapFor(extensionSet.Extensions, [&] (TFluentMap fluent, const TExtension& extension) {
@@ -433,7 +433,8 @@ void Serialize(const TExtensionSet& extensionSet, NYson::IYsonConsumer* consumer
                     ParseProtobuf(
                         fluent.GetConsumer(),
                         &inputStream,
-                        ReflectProtobufMessageType(extensionDescriptor->MessageDescriptor));
+                        ReflectProtobufMessageType(extensionDescriptor->MessageDescriptor),
+                        parserOptions);
                 });
         });
 }
@@ -463,7 +464,7 @@ void Deserialize(TExtensionSet& extensionSet, NYTree::INodePtr node)
     }
 }
 
-REGISTER_INTERMEDIATE_PROTO_INTEROP_REPRESENTATION(NYT::NProto::TExtensionSet, TExtensionSet)
+REGISTER_INTERMEDIATE_PROTO_INTEROP_REPRESENTATION_WITH_OPTIONS(NYT::NProto::TExtensionSet, TExtensionSet)
 
 ////////////////////////////////////////////////////////////////////////////////
 
diff --git a/yt/yt/core/yson/protobuf_interop-inl.h b/yt/yt/core/yson/protobuf_interop-inl.h
index 50ec342fe4..af20635494 100644
--- a/yt/yt/core/yson/protobuf_interop-inl.h
+++ b/yt/yt/core/yson/protobuf_interop-inl.h
@@ -17,6 +17,40 @@ const TProtobufMessageType* ReflectProtobufMessageType()
 
 ////////////////////////////////////////////////////////////////////////////////
 
+template <class ProtoType, class Type, bool UseParseOptionsInSerialize>
+static const void* DoRegisterIntermediateProtoInteropRepresentation() {
+    NYT::NYson::AddProtobufConverterRegisterAction([] {
+        auto* descriptor = ProtoType::default_instance().GetDescriptor();
+        NYT::NYson::TProtobufMessageConverter converter;
+        converter.Serializer = [] (
+            NYT::NYson::IYsonConsumer* consumer,
+            const google::protobuf::Message* message,
+            const NYson::TProtobufParserOptions& parseOptions = {})
+            {
+                const auto* typedMessage = dynamic_cast<const ProtoType*>(message);
+                YT_VERIFY(typedMessage);
+                Type value;
+                FromProto(&value, *typedMessage);
+                if constexpr (UseParseOptionsInSerialize) {
+                    Serialize(value, consumer, parseOptions);
+                } else {
+                    Serialize(value, consumer);
+                }
+            };
+        converter.Deserializer = [] (google::protobuf::Message* message, const NYT::NYTree::INodePtr& node) {
+            auto* typedMessage = dynamic_cast<ProtoType*>(message);
+            YT_VERIFY(typedMessage);
+            Type value;
+            Deserialize(value, node);
+            ToProto(typedMessage, value);
+        };
+        NYT::NYson::RegisterCustomProtobufConverter(descriptor, converter);
+    });
+    return nullptr;
+};
+
+////////////////////////////////////////////////////////////////////////////////
+
 std::optional<int> FindProtobufEnumValueByLiteralUntyped(
     const TProtobufEnumType* type,
     TStringBuf literal);
diff --git a/yt/yt/core/yson/protobuf_interop.cpp b/yt/yt/core/yson/protobuf_interop.cpp
index d6952ed4f5..c23dcab2a3 100644
--- a/yt/yt/core/yson/protobuf_interop.cpp
+++ b/yt/yt/core/yson/protobuf_interop.cpp
@@ -2693,7 +2693,7 @@ private:
                             PooledString_.resize(length);
                             CodedStream_.ReadRaw(PooledString_.data(), PooledString_.size());
                             Y_UNUSED(message->ParseFromArray(PooledString_.data(), PooledString_.size()));
-                            converter.Serializer(Consumer_, message.get());
+                            converter.Serializer(Consumer_, message.get(), Options_);
                             YPathStack_.Pop();
                         } else {
                             LimitStack_.push_back(CodedStream_.PushLimit(static_cast<int>(length)));
diff --git a/yt/yt/core/yson/protobuf_interop.h b/yt/yt/core/yson/protobuf_interop.h
index 3e6faa6a53..bdc06244da 100644
--- a/yt/yt/core/yson/protobuf_interop.h
+++ b/yt/yt/core/yson/protobuf_interop.h
@@ -198,7 +198,7 @@ void AddProtobufConverterRegisterAction(std::function<void()> action);
 
 struct TProtobufMessageConverter
 {
-    std::function<void(IYsonConsumer* consumer, const google::protobuf::Message* message)> Serializer;
+    std::function<void(IYsonConsumer* consumer, const google::protobuf::Message* message, const TProtobufParserOptions& parserOptions)> Serializer;
     std::function<void(google::protobuf::Message* message, const NYTree::INodePtr& node)> Deserializer;
 };
 
@@ -207,29 +207,13 @@ void RegisterCustomProtobufConverter(
     const google::protobuf::Descriptor* descriptor,
     const TProtobufMessageConverter& converter);
 
-#define REGISTER_INTERMEDIATE_PROTO_INTEROP_REPRESENTATION(ProtoType, Type)                                             \
-    YT_ATTRIBUTE_USED static const void* PP_ANONYMOUS_VARIABLE(RegisterIntermediateProtoInteropRepresentation) = [] {   \
-        NYT::NYson::AddProtobufConverterRegisterAction([] {                                                             \
-            auto* descriptor = ProtoType::default_instance().GetDescriptor();                                           \
-            NYT::NYson::TProtobufMessageConverter converter;                                                            \
-            converter.Serializer = [] (NYT::NYson::IYsonConsumer* consumer, const google::protobuf::Message* message) { \
-                const auto* typedMessage = dynamic_cast<const ProtoType*>(message);                                     \
-                YT_VERIFY(typedMessage);                                                                                \
-                Type value;                                                                                             \
-                FromProto(&value, *typedMessage);                                                                       \
-                Serialize(value, consumer);                                                                             \
-            };                                                                                                          \
-            converter.Deserializer = [] (google::protobuf::Message* message, const NYT::NYTree::INodePtr& node) {       \
-                auto* typedMessage = dynamic_cast<ProtoType*>(message);                                                 \
-                YT_VERIFY(typedMessage);                                                                                \
-                Type value;                                                                                             \
-                Deserialize(value, node);                                                                               \
-                ToProto(typedMessage, value);                                                                           \
-            };                                                                                                          \
-            NYT::NYson::RegisterCustomProtobufConverter(descriptor, converter);                                         \
-        });                                                                                                             \
-        return nullptr;                                                                                                 \
-    } ();
+#define REGISTER_INTERMEDIATE_PROTO_INTEROP_REPRESENTATION(ProtoType, Type)                        \
+    YT_ATTRIBUTE_USED static const void* PP_ANONYMOUS_VARIABLE(RegisterIntermediateProtoInteropRepresentation) = \
+        NYson::DoRegisterIntermediateProtoInteropRepresentation<ProtoType, Type, false>();
+
+#define REGISTER_INTERMEDIATE_PROTO_INTEROP_REPRESENTATION_WITH_OPTIONS(ProtoType, Type)            \
+    YT_ATTRIBUTE_USED static const void* PP_ANONYMOUS_VARIABLE(RegisterIntermediateProtoInteropRepresentationWithOptions) =  \
+        NYson::DoRegisterIntermediateProtoInteropRepresentation<ProtoType, Type, true>();
 
 ////////////////////////////////////////////////////////////////////////////////
 
diff --git a/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp b/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp
index e9c4a9c0b8..7373b8d78c 100644
--- a/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp
+++ b/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp
@@ -2036,6 +2036,27 @@ TEST(TProtobufToYsonTest, UnknownFields)
     }
 }
 
+TEST(TProtobufToYsonTest, UnknownFieldsNested)
+{
+    TProtobufParserOptions options;
+    options.SkipUnknownFields = true;
+
+    TEST_PROLOGUE()
+    codedStream.WriteTag(WireFormatLite::MakeTag(15 /*nested_message1*/, WireFormatLite::WIRETYPE_LENGTH_DELIMITED));
+    codedStream.WriteVarint64(3);
+    codedStream.WriteTag(WireFormatLite::MakeTag(42 /*unknown*/, WireFormatLite::WIRETYPE_VARINT));
+    codedStream.WriteVarint64(2);
+    TEST_EPILOGUE_WITH_OPTIONS(TMessage, options)
+
+    NProto::TMessage message;
+    TryDeserializeProto(&message, TRef::FromString(protobuf));
+
+    TString newYsonString;
+    TStringOutput newYsonOutputStream(newYsonString);
+    TYsonWriter ysonWriter(&newYsonOutputStream, EYsonFormat::Pretty);
+    WriteProtobufMessage(&ysonWriter, message, options);
+}
+
 TEST(TProtobufToYsonTest, ReservedFields)
 {
     TEST_PROLOGUE()